BEIJING, China
Small shipments reach leading Chinese technology groups as Beijing balances demand for advanced AI computing with support for domestic semiconductor development
Chinese authorities have begun permitting limited shipments of Nvidia’s H200 artificial intelligence processors into mainland China, marking an easing of restrictions as the country’s leading technology companies compete to develop increasingly advanced AI models.
According to reporting by the Financial Times, ByteDance and Tencent have each received approximately 10,000 H200 processors in recent weeks, while several other Chinese technology groups could receive approval for shipments of a similar size. Reuters subsequently reported the development, citing the Financial Times, but said it had not independently verified the reported shipments.
The development represents a notable shift after months of uncertainty surrounding access to Nvidia’s advanced AI hardware. Chinese technology companies have sought greater computing capacity for the training of increasingly sophisticated artificial intelligence models, while Beijing has simultaneously pursued policies intended to strengthen the country’s domestic semiconductor industry.
Under U.S. rules introduced earlier this year, exports of Nvidia H200 processors and comparable advanced chips to approved customers in China may be considered on a case-by-case basis, subject to security, compliance and testing requirements. The U.S. Department of Commerce’s Bureau of Industry and Security formally revised its licensing policy for the H200 in January 2026.
Limited Mainland Access Remains
Despite the latest easing, access to the H200 within mainland China remains controlled.
The Financial Times reported that Chinese companies cleared by U.S. authorities to purchase considerably larger quantities of H200 processors are being encouraged by Beijing to keep much of that computing capacity outside the mainland. Chinese regulators have also told companies that H200 processors may be deployed in Hong Kong, which operates outside mainland China’s customs territory.
The approach could allow Chinese AI developers to gain access to additional Nvidia computing power while Beijing continues promoting domestically produced alternatives.
Purchases involving H200-powered systems are also reported to require approval through an application process overseen by China’s National Development and Reform Commission.
AI Developers Seek Greater Computing Capacity
Advanced processors have become an increasingly important component of the international competition to develop large-scale artificial intelligence systems.
Chinese technology companies and AI laboratories are investing heavily in model development as they seek to narrow the performance gap with leading U.S. developers. Access to high-performance GPUs is particularly important during AI model training, which can require large clusters of processors operating for extended periods.
Nvidia describes the H200 as a Hopper-architecture Tensor Core GPU designed for generative AI and high-performance computing. The processor incorporates 141GB of HBM3e memory and memory bandwidth of 4.8 terabytes per second, providing substantial computing capacity for large language models and other data-intensive workloads.
Although Nvidia has since introduced newer generations of AI hardware, restrictions on the export of its most advanced processors continue to limit the products available to customers in China.
China Continues Push for Domestic Semiconductor Development
The measured relaxation of H200 restrictions highlights the competing priorities confronting Chinese policymakers.
Beijing has sought to reduce dependence on foreign semiconductor technology and expand the use of domestically developed processors, including products from Chinese chipmakers. At the same time, growing demand for computing infrastructure from the country’s AI industry has created pressure for additional supplies of high-performance hardware.
The latest development suggests Chinese authorities may continue permitting controlled access to foreign processors where necessary for advanced AI development while maintaining broader policies designed to strengthen domestic semiconductor capabilities.
The Financial Times reported that Nvidia had approximately 500,000 H200 processors in inventory primarily intended for Chinese customers and that some Nvidia partners had recently resumed accepting orders for systems incorporating the processors, although individual purchases remain subject to regulatory approval.
Nvidia, ByteDance, Tencent and China’s National Development and Reform Commission had not provided comments to the Financial Times regarding the reported policy changes at the time of publication.
Industry Outlook
The movement of H200 processors into China adds another development to the evolving technology relationship between Washington and Beijing, where access to advanced semiconductor technology has become a central economic and national-security issue.
Further changes to Chinese import approvals, U.S. export licensing policies and the development capacity of domestic Chinese chipmakers are expected to remain significant factors shaping the availability of computing infrastructure for the country’s rapidly expanding artificial intelligence sector.
Media Contact Details
Nvidia
Email: Send Email
